Recommender systems are commonly used for recommending items such as products, restaurants or other points-ofinterest (POI). In our automotive scenario, the driver of a car gets recommendations for gas stations. Thereby, item attributes such as price or location are important, but also context data such as the current time, location or gas level of the car when requesting the recommendation. Our approach is based on Multi-Criteria Decision Making (MCDM) methods to calculate scores on several dimensions. We used utility functions modeling the importance of different route context elements which were derived from a preliminary user study, among other information. In addition, our system performs contextual pre-and post-filtering to reduce the number of considered items. The evaluation showed that our approach produced reasonably good results in comparison with the assessment of users in a second study.
Abstract. Service-orientation enables dynamic interoperation of distributed services and facilitates seamless service provision or runtime creation of new applications. This dynamic service composition is particularly powerful in peer-to-peer (P2P) systems which offer scalability through self-management and autonomy. However, P2P service composition is nontrivial due to permanent peer churn and lack of central control. Existing approaches reduce composite service initialization to an NPhard path finding problem. Thus, peer failure adaptation is costly and runtime consideration of peer logons or load changes is not practicable. This paper introduces logical peer groups for service composition. They enable runtime composite service reconfiguration including the migration of services to other peers. A prototype implementation is presented and the algorithms are evaluated through both formal and empirical analysis. The evaluation shows that the approach results in significant reduction of computational complexity, improves fault-tolerance and enables adaptation of logons and load changes which has not been possible so far.
This chapter presents an approach to extend a real world mobile tourist guide running on personal digital assistants (PDAs) with collaborative filtering. The system builds a model of item similarities based on explicit and implicit ratings. This model is then utilized to generate recommendations in several ways. The approach integrates the current user location as context. Experiences gained in two field studies are reported. In the first one, 30 participants – real tourists visiting Prague – used the recommender function and were asked to fill out a questionnaire with promising results. In a second field study analyzing usage log files, an improvement of recommendations based on the collaborative filter in comparison to the pure location-based filter used before was discovered. In addition, recommendations based on implicit ratings derived from audio playback duration outperformed the model based on explicit ratings.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.